Up to this point we have been dealing with the seed in 

 its coats. But if we remove these coverings we find another 

 singular contrast in the behaviour of these, two seeds. In 

 the instance of Canavalia ensiformis we discover that it makes 

 no essential difference whether we employ the seed in its 

 coverings, or puncture it through its coats, or deprive it of 

 them altogether. In any case the same average weight is 

 maintained, the baring of the kernel or puncturing of the 

 coats merely resulting in a small increase of the hygroscopic 

 range. Whatever may be the function of the coats of a 

 permeable seed, they do not prevent it from responding from 

 day to day to the variations in the atmospheric humidity, 

 though they may regulate the process. This would seem 

 to be true of the large majority of similar seeds, and it 

 follows naturally from the permeable character and the 

 hygrometric behaviour of the seed-coverings. (See Note 6 

 of the Appendix.) 



In the table below I have given the results of simultaneous 

 observations on the bared, punctured, and entire seeds of 

 Canavalia ensiformis collected from the plant at the same time. 

 The table explains itself, except that one may add that the 

 hygroscopic or hygrometric variation is the range of the 

 changes in weight exhibited in the course of two or three 

 weeks stated as a percentage of the total weight. 



COMPARISON OF THE BEHAVIOUR OF THE SEEDS OF CANAVALIA 

 ENSIFORMIS WHEN BARED, PUNCTURED, AND ENTIRE. 



The influence 

 of the seed- 

 coats on the 

 changes in 

 the seed's 

 weight, 



(a) in the 

 case of a 

 permeable 

 seed, 



If we repeat these experiments with the seeds of Guilandma (b) in the 

 bonducella^ and either remove the hard, impermeable, shell-like impermeable 

 coats or pierce them with a file, we obtain results of quite a seed - 



